DOI QR코드

DOI QR Code

An evaluation methodology for cement concrete lining crack segmentation deep learning model

콘크리트 라이닝 균열 분할 딥러닝 모델 평가 방법

  • Ham, Sangwoo (Dept. of Geoinformatics, University of Seoul) ;
  • Bae, Soohyeon (Dept. of Geoinformatics, University of Seoul) ;
  • Lee, Impyeong (Dept. of Geoinformatics, University of Seoul) ;
  • Lee, Gyu-Phil (Dept. of Geotechnical Engineering Research, Korea Institute of Civil Engineering and Building Technology) ;
  • Kim, Donggyou (Dept. of Geotechnical Engineering Research, Korea Institute of Civil Engineering and Building Technology)
  • 함상우 (서울시립대학교 대학원 공간정보공학과) ;
  • 배수현 (서울시립대학교 대학원 공간정보공학과) ;
  • 이임평 (서울시립대학교 공간정보공학과) ;
  • 이규필 (한국건설기술연구원 지반연구본부) ;
  • 김동규 (한국건설기술연구원 지반연구본부)
  • Received : 2022.09.26
  • Accepted : 2022.10.18
  • Published : 2022.11.30

Abstract

Recently, detecting damages of civil infrastructures from digital images using deep learning technology became a very popular research topic. In order to adapt those methodologies to the field, it is essential to explain robustness of deep learning models. Our research points out that the existing pixel-based deep learning model evaluation metrics are not sufficient for detecting cracks since cracks have linear appearance, and proposes a new evaluation methodology to explain crack segmentation deep learning model more rationally. Specifically, we design, implement and validate a methodology to generate tolerance buffer alongside skeletonized ground truth data and prediction results to consider overall similarity of topology of the ground truth and the prediction rather than pixel-wise accuracy. We could overcome over-estimation or under-estimation problem of crack segmentation model evaluation through using our methodology, and we expect that our methodology can explain crack segmentation deep learning models better.

터널을 비롯한 여러 가지 기반시설물에 발생한 콘크리트 균열을 영상과 딥러닝 기반으로 자동 탐지하는 연구가 최근 활발히 이루어지고 있다. 이러한 연구성과를 실제 현장에 적용하려면 딥러닝 모델의 신뢰성을 설명할 수 있어야한다. 본 연구에서는 선형성이 강한 균열의 기하적인 특성을 고려했을 때 화소 기반으로 계산하는 기존 평가지표가 충분치 않다는 점을 지적하며, 균열 분할 딥러닝 모델의 성능을 더 합리적으로 설명할 수 있는 다른 평가지표를 제시하고 비교 분석한다. 먼저 선형 객체의 유사성을 측정할 수 평가방법을 제시한다. 구체적으로는 기준 데이터에 허용 버퍼(tolerance buffer)를 부여하여 평가하는 방법을 설계, 구현, 검증한다. 실험 결과 본 연구에서 제안하는 방법은 균열 분할 딥러닝 모델 평가시 기존 대비 과대평가 또는 과소평가 문제를 해결할 수 있었으며, 화소 기반 성능 평가 지표에 비해 균열 분할 딥러닝 모델의 성능을 더 잘 설명할 것으로 기대한다.

Keywords

Acknowledgement

본 논문은 한국건설기술연구원 주요사업으로 지원을 받아 수행된 연구(인공지능을 활용한 대심도 지하 대공간의 스마트 복합 솔루션 개발)로 이에 감사합니다.

References

  1. Bianchi, E., Abbott, A.L., Tokekar, P., Hebdon, M. (2021), "COCO-Bridge: structural detail data set for bridge inspections", Journal of Computing in Civil Engineering, Vol. 35, No. 3, 04021003.
  2. Chen, L.C., Zhu, Y., Papandreou, G., Schroff, F., Adam, H. (2018), "Encoder-decoder with atrous separable convolution for semantic image segmentation", arXiv:1802:02611, pp. 1-23.
  3. Garcia-Garcia, A., Orts-Escolano, S., Oprea, S.O., Villena-Martinez, V., Garcia-Rodriguez, J. (2017), "A review on deep learning techniques applied to semantic segmentation", arXiv:1704:06857, pp. 1-22.
  4. Ham, S., Bae, S., Kim, H., Lee, I., Lee, G.P., Kim, D. (2021), "Training a semantic segmentation model for cracks in the concrete lining of tunnel", Journal of Korean Tunnelling and Underground Space Association, Vol. 23, No. 6, pp. 549-558. https://doi.org/10.9711/KTAJ.2021.23.6.549
  5. Hsieh, Y.A., Tsai, Y.J. (2020), "Machine learning for crack detection: review and model performance comparison", Journal of Computing in Civil Engineering, Vol. 34, No. 5, 04020038.
  6. Kim, A.R., Kim, D., Byun, Y.S., Lee, S.W. (2018), "Crack detection of concrete structure using deep learning and image processing method in geotechnical engineering", Journal of the Korean Geotechnical Society, Vol. 34, No. 12, pp. 145-154. https://doi.org/10.7843/KGS.2018.34.12.145
  7. Liu, Y., Yao, J., Lu, X., Xie, R., Li, L. (2019), "DeepCrack: A deep hierarchical feature learning architecture for crack segmentation", Neurocomputing, Vol. 338, pp. 139-153. https://doi.org/10.1016/j.neucom.2019.01.036
  8. Middha, L. (2020), "Crack segmentation dataset", https://www.kaggle.com/lakshaymiddha/crack-segmen tation-dataset (September 20, 2022).
  9. MMSegmentation Contributors (2020), "OpenMMLab semantic segmentation toolbox and benchmark", https://github.com/open-mmlab/mmsegmentation (September 20, 2022).
  10. Mundt, M., Majumder, S., Murali, S., Panetsos, P., Ramesh, V. (2019), "Meta-learning convolutional neural architectures for multi-target concrete defect classification with the concrate defect bridge image dataset", Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, California, pp. 11196-11205.
  11. OurBrain (2022), "Big Eye System", OurBrain, http://ourbrain.co.kr/ (September 20, 2022).
  12. Paik, S.H., Choi, D.Y., Kim, Y.K., Jung, S.W., Kim, D.N. (2021), "Implementation of the drones with deep-learning crack detection analysis for inspection of bridge", The Journal of Korean Institute of Information Technology, Vol. 19, No. 3, pp. 45-52.
  13. Park, G. (2022), "Building crack detection images", AI Hub, https://aihub.or.kr/aihubdata/data/view.do?currMenu=115&topMenu=100&aihubDataSe=realm&dataSetSn=162 (September 20, 2022).
  14. Scikit-image contributors (2022), "Morphological filtering", Scikit-image, https://scikit-image.org/docs/stable/auto_examples/applications/plot_morphology.html (September 20, 2022).
  15. Tsai, Y.C., Chatterjee, A. (2017), "Comprehensive, quantitative crack detection algorithm performance evaluation system", Journal of Computing in Civil Engineering, Vol. 31, No. 5, 04017047.
  16. Zhang, T.Y., Suen, C.Y. (1984), "A fast parallel algorithm for thinning digital patterns", Communications of the ACM, Vol. 27, No. 3, pp.236-239. https://doi.org/10.1145/357994.358023